One of the most challenging aspects of threatened species recovery planning is understanding the drivers of population declines over broad spatial extents, especially where threats vary across a species’ range. This is certainly true for the koala; a species that is widely distributed across eastern Australia, but where the the threats driving declines vary substatially depending on where you are. The main threats include habitat loss and urbanisation, drought and climate change, dog attacks, vehicle collisions and disease. Undertandably, for such a widely distributed species, the nature of these threats varies markedly across the koala’s range (e.g., urbanisaiton is more prevalent in the east of the koala’s range and drought is more prevalent in the west). The main challenge to understand the impact of these threats spatially for koalas, therefore, is obtaining data on threats and the response of koalas to those threats across broad spatial extents.
The type of data needed is expensive to collect and logistically difficult to obtain. But this is where citizen science can make a big difference. In a recent paper of ours in Diversity and Distributions by Truly Santika and others, we use citizen science data on koala distributions since 1987 for New South Wales, Australia to quantify the implications of four threats (habitat loss, urbanisation, roads and climate change) acorss the state (see the paper here). In a novel step we combine surveys of the public on whether they had seen koalas where they live with a dynamic occupancy model that accounts for extinction and colonisation dynamics and detection error. Importantly, we were able to construct presence / absence data and account for detection errors by also using data on common species that the public had also seen. Where the public had seen common species we assumed that they were actively looking for wildlife and a failure to sight koalas was therefore counted as an absence. This consideration of data on other species was therefore a major factor that allowed us to fit a dynamic occupancy model to the data.
Our model predicts higher extinction risk in the western parts of New South Wales than the eastern parts and identifies the major drivers of extinction risk as being habitat loss, it’s interaction with urban density and temperature. Rainfall and road density were less important. Importantly, the model allows for spatial variation in the effect of these threats across the state to be estimated. Recently there has been increasing recognition that citizen science can make a crucial contribution to biodiversity conservaiton (Silvertown 2010, Tulloch et al. 2013). Our study is another important example of the power of citizen science data that is enhanced by using state-of-the-art statistical techniques.
To read a recent Decision Point article on this paper see here.